Pasha Bolokhov <> writes:

>>> +             test_cmp status.actual.2 status.expect.2
>> It is customary to call the files 'expect' and 'actual'. Furthermore,
>> swap the order so that in case of a failure the diff shows how the
>> actual text was changed from the expected text:
>>                 test_cmp status.expect.2 status.actual.2
> So, is naming the files "status.expect.2" instead of just
> "expect"/"actual" ok or not?

Unless there is a compelling reason not to, just use expect/actual,
without "status." or ".<number>".  When debugging, people pass the
"-i" (and often "-v") to the script which would stop at the first
breakage.  If they can inspect the result in "actual", that is far
more pleasant than having to know that it is status.actual.43 they
have to inspect while ignoring clutters in status.actual.{1-42} that
have nothing to do with the breakage they are dealing with.

> Ok, no problem, will change that. The only thing, I saw this in other
> tests, so decided to use it too. Those tests use "test_mighf_fail rm"

"might_fail" and "must_fail" are totally different beasts.
"must_fail" is about controlled failure exit (i.e. no segfaulting
allowed) and we avoid using it on non-git things.

When you run clean-up commands without knowing if there is anythning
to clean-up, e.g. you may or may not have successfully created a
junk file and you want to make sure that file does not exist.  It
happens that "rm junk" has a convenient option "-f" to make it not
barf upon missing "junk", so you can do so with "rm -f junk" instead
of saying "test_might_fail rm junk", but "might_fail" is designed to
be useful for commands that do not have such a convenient option.
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to
More majordomo info at

Reply via email to